DataLoader Node

The DataLoader node is the starting point for most pipelines. It loads data files and passes them to downstream nodes for processing.

Overview

Property	Value
Type	Source node
Inputs	None
Outputs	DataFrame
Supported formats	CSV, Parquet

Configuration

File Path

The path to your data file. Click Browse to select a file, or enter the path manually.

/Users/yourname/Desktop/data.csv

Supported Formats

Format	Extension	Notes
CSV	.csv	Comma-separated values
Parquet	.parquet	Columnar storage format

Tip

Parquet files load faster and use less memory for large datasets.

Output

The DataLoader outputs a pandas DataFrame with:

• All columns from the source file
• Inferred data types (numeric, string, datetime)
• Original row order preserved

Data Preview

After selecting a file, the node shows a preview:

• Rows: Total number of rows
• Columns: Column names and types
• Sample: First 5 rows of data

Usage Examples

Basic Loading

1. Add a DataLoader node to the canvas
2. Click the node to open the properties panel
3. Click Browse and select your CSV file
4. Connect to a Trainer or Script node

Large Files

For files over 100MB:

1. Consider converting to Parquet format first
2. Use a Script node to sample or filter data
3. Check available memory before loading

Common Issues

”File not found”

• Check that the file path is correct
• Ensure the file hasn’t been moved or renamed
• Use absolute paths (starting with /)

“Encoding error”

• CSV files should use UTF-8 encoding
• If your file uses a different encoding, convert it first:

iconv -f ISO-8859-1 -t UTF-8 input.csv > output.csv

“Memory error”

• File is too large for available RAM
• Sample the data or use a smaller subset
• Consider using Parquet format

Generated Code

When the pipeline runs, DataLoader generates:

import pandas as pd

# Load the data
df = pd.read_csv("/path/to/data.csv")

# Display info
print(f"Loaded {len(df)} rows, {len(df.columns)} columns")
print(df.dtypes)

Best Practices

1. Use descriptive file names — customer_churn_2024.csv not data.csv
2. Keep data files in a consistent location — Avoid moving files after creating pipelines
3. Check data quality — Preview the data to ensure it loaded correctly
4. Use Parquet for large files — Faster loading and smaller file size

Related Nodes

• Trainer — Train models on loaded data
• Script — Custom data preprocessing